Working Groups‎ > ‎

WG5: Applications & Performance Evaluation

WG Leader: Prof. Pascal Felber

TMs appear to have a huge potential in simplifying the development of parallel applications. On the other hand, being the research field on TMs still in its infancy, the number of complex benchmarks for TMs currently available is still very limited, and the real-world applications pioneering the adoption of TMs are probably even less. This represents a major impairment not only for realistically evaluating the performance of the various TM solutions proposed in literature, but also for precisely assessing the usability of TMs in complex, large scale applications and across the many different application domains that could potentially benefit from their adoption (including web-based applications, video-games, CAD systems, stream processors, financial, healthcare, video-editing, navigation systems, simulators, graph analysis toolkits, just to mention a few).

Leveraging on the research results developed by WG2, WG3 and WG4, this WG will unite researchers and ICT practioners working on the development of TMs and/or TM-based applications. This will permit, on one hand, to ensure timely interactions among developers of TM-based applications with the purpose of establishing and disseminate best-practices; on the other hand, it will create bi-directional communication channels between developers of TM-based applications and developers of TMs, to assist the former ones in the development and tuning of applications and provide the latter ones with valuable feedback from real-world settings. Overall, this will bring the twofold benefit of fostering the adoption of TMs in the software industry, and of driving future research efforts to actually meet the requirements of real-world applications.

An important research goal of this WG is also to establish automatized methods (based on analytical models/simulations or both) assisting the application's developers in the choice of the TM implementation maximizing the performance of their own application, forecasting the scale-up achievable by acquiring additional or more powerful hardware (e.g. in a cloud-computing settings), and identifying the performance bottlenecks both at the application level and TM level.